Back in June 2012, my paper "Galois cohomology of a number field is Koszul" was rejected by your journal based on a referee's report stating that
"This paper is not written to be read by anyone not very familiar with the previous work       of the author. In addition, it seems strange that this paper was largely done in 1995 but that the exposition has not shown the kind of improvement one would expect over 17 years. I spent hours on the first 8 pages, and refuse to read further until the author rewrites the paper for an audience which includes me."
I, for one, always thought that a good command of the paper's subject was a requirement for a reviewer, and considered spending hours to be a matter of course. I also tend to approach the refereeing as a kind of almost-research work, taking it to be the reviewer' main task to understand and explain what is written in the paper under review (rather than judge the paper's quality based on considerations of publishing capacity and backlog).
An editorial process satisfied with making its decisions based on the kind of refereeing job exemplified above has no use for my humble services.